March 24, 2026
Receipts or it didn’t ship
Show HN: ProofShot – Give AI coding agents eyes to verify the UI they build
AI code gets receipts — fans cheer, skeptics yell “just use Playwright”
TLDR: ProofShot records an AI assistant’s build session and packages video, screenshots, and errors so you can verify what happened. Commenters are split: some say it’s useful proof, others argue rivals like exe.dev and Microsoft’s Playwright already cover this, while debating true “vision,” desktop support, and deeper browser control.
ProofShot promises to give your robot coder “eyes,” recording every click and error while it builds your app and bundling it into video and screenshots so you’ve got receipts. The pitch is simple: spin up your app, open a hidden browser, capture everything, trim the boring bits, and export proof. The community? Oh, they had thoughts.
One early voice pulled the rug: “exe.dev already does this,” joked a user while hinting for a sponsorship, cheering that at least this is open source. Another commenter went full alphabet soup with “MCP” (a way for chatbots to use tools) and “CDP” (a way to control Chrome), bragging their Claude bot drives Brave like a self-driving car — even the UI. Meanwhile, the practical crowd demanded: does it handle desktop apps or just websites? Then came the spicy philosophy: “Screenshots aren’t ‘seeing’.” In other words, recording isn’t intelligence — the real win would be auto-spotting issues and suggesting fixes.
And the heavyweight comparison dropped like a mic: what does this do that Playwright CLI doesn’t? Cue meme energy: “Nice receipts” vs “another wrapper for headless Chrome,” plus “can’t wait to watch my bot fail in 4K.” The verdict: excitement for proof-of-work, but a split over whether this is fresh innovation or a slick remix of tools devs already use.
Key Points
- •ProofShot wraps a development server to record AI coding agents’ actions and outputs a proof bundle.
- •It launches the dev server, opens a headless Chromium session, and records video of the process.
- •The tool collects errors, stops recording, trims idle time, and generates proof artifacts.
- •Outputs are designed to provide concise evidence of the agent’s development workflow.
- •Setup is via a CLI with a quick-start process advertised to take about 30 seconds.